HVAC
HVAC AI Answering: The Summit Air Worked Example
An explicitly illustrative six-truck scenario for testing call coverage, booking, follow-up, membership workflows, and defensible measurement.

Summit Air & Heat and Dan K. are disclosed illustrative stand-ins, not an identified customer or a real quoted owner. The amounts and counts below are scenario inputs—not measured LumiTalk results, typical outcomes, or a forecast. The value of the example is the measurement design: replace every input with permissioned source records before presenting it as a customer case study.
The problem (March)
March was the warning shot. Two CSRs can only hold two lines. When a third call rings, it goes to voicemail. In April, in Bakersfield, that is a problem.
- 71% of calls answered. Almost one in three callers hit voicemail or a busy signal.
- 26 hours of CSR overtime just to keep up with what did get through.
- No idea how much work walked out the door, because you can’t measure the call you never picked up.
The heat wave (April 14)
April 14 hit 104 degrees. In March, that week buries two CSRs and sends the overflow to voicemail. This time the overflow had somewhere to go. Lumi held every extra line, booked the emergencies, and never once put a sweating homeowner on hold. 121 of the month’s calls came in at night or on a weekend. Every one got answered live.
Illustrative scenario inputs
- The scenario assumes 412 calls offered and 412 completed, including 121 nights and weekends; a real report must include abandoned, failed, overflowed, and transferred calls.
- It models 96 booked jobs and $61,400 in booked revenue; neither the bookings nor ServiceTitan write operations are verified customer results.
- It models three revived replacement estimates worth $23,400 after follow-up; a real study must disclose attribution, contact history, financing, cancellations, and collected revenue.
- It models 41 membership renewals; a real study must define eligible memberships, contact and consent policy, renewal window, and net retained revenue.
- The $84,800 total is illustrative booked value, not verified collected revenue, gross profit, or return on investment.
I stopped losing the coin flip on every third call. The heat wave used to break us. This April it just filled the board. I paid for Lumi in the first week and never looked back.
— Dan K., Owner, Summit Air & Heat
A note on methodology: Summit Air & Heat is an illustrative composite, not a named customer, and Dan K. is not a real quoted person. The figures are scenario inputs selected to make the measurement worksheet concrete; they are not typical results, a forecast, a testimonial, or evidence of what a six-truck shop should expect. A future customer case study would require permission and source records.
What a real HVAC case study must prove
- Customer identity, publication permission, business size, market, season, and measurement period.
- Pre-change baseline from the same source systems and definitions for calls offered, answered, qualified, booked, completed, canceled, and collected.
- Deployment date, covered hours and channels, scripts, escalation rules, software relationship, enabled operations, and human work retained.
- Source-level attribution for booking, follow-up, membership, and revenue outcomes, including overlaps and exclusions.
- Vendor fees, internal labor, discounts, refunds, fulfillment cost, and whether results are booked revenue, collected revenue, contribution margin, or gross profit.
- Confounding events such as weather, advertising spend, staffing, pricing, service-area changes, or promotions.
Use the scenario as a measurement worksheet
Export a baseline period and a comparable post-change period. Preserve source identifiers so the same call cannot be credited once as an after-hours booking and again as a revived estimate without disclosure. Google Ads' official call-conversion documentation illustrates the identifiers, timestamps, consent, and duplicate handling needed when an advertising call is reconciled to a later outcome. Report denominators, not just wins: calls offered, eligible calls, successful records, confirmed appointments, completed jobs, collected value, costs, and exception rates. Compare against the same season where possible and explain any unavoidable difference.
Continue through the HVAC evaluation cluster
Use the winter worked example for safety and after-hours edge cases, the cost guide for the economic model, and the ServiceTitan article for the field-level integration contract. winter after-hours worked example · HVAC answering-service cost guide · ServiceTitan booking workflow · LumiTalk for HVAC
Run these numbers on your own call volume.
See Lumi for HVACQuick answers
Frequently asked
How fast do HVAC shops see results with an AI answering service?
This worked example does not establish a time to results. Measure from the actual activation date and report calls offered, completed intakes, confirmed bookings, completed jobs, collected value, costs, exceptions, and a comparable baseline. The $84,800 figure here is an illustrative input, not a customer result or ROI claim.
Does it work during a heat-wave call spike?
A heat wave is a useful concurrency and recovery test. The 412-of-412 figure in this worked example is a scenario assumption, not a measured LumiTalk result. A real test should report documented concurrency, queue or overflow behavior, completed and abandoned calls, transfers, booking confirmation, exceptions, and recovery.
Are these real numbers?
No. This is an illustrative composite and measurement worksheet, not a customer case study. Summit Air & Heat and Dan K. are stand-ins, and the figures are hypothetical inputs—not typical results, a forecast, or measured performance.
See Lumi answer HVAC calls 24/7
Watch Lumi triage a no-cool call, agree the trip fee, and book it straight into ServiceTitan — then run your own numbers on the leak calculator.








